A number of vascular diseases and vascular risk factors including diabetes, hypertension, hyperlipidemia, smoking, and obesity have been implicated but not consistently established as risk factors for Alzheimer's disease (AD). In addition, studies using a combination of these risk factors to predict AD risk have reported only modest accuracy. Current predictive models for AD have typically characterized risk exposure by assessing vascular markers at a single point in time at the baseline. Such characterization fails to capture potential changes or variability over the relatively long latency period prior to he onset of AD symptoms. These static predictive models also ignore the vast heterogeneity in individuals'longitudinal vascular markers over time. We propose a secondary data analysis developing dynamic models using longitudinally collected vascular markers to predict AD risk. We will merge electronic medical records of participants enrolled in the Indianapolis cohort of the longitudinal community-based Indianapolis-Ibadan Dementia Project (IIDP) with research data collected in the IIDP. The IIDP has enrolled a total of 4,105 African Americans aged 65 or older and followed the participants for up to 19 years with cognitive evaluation, clinical diagnosi and risk factor information at regularly scheduled intervals every 2 to 3 years. Our analyses will focus on longitudinally measured vascular markers including blood pressure, lipids, hemoglobin A1C and fasting glucose levels obtained from electronic medical records for the risk of AD.
In Aim 1, we will compare longitudinal vascular risk factor profiles between participants with AD and those with normal cognition and determine whether differences in longitudinal vascular profiles are accounted for by differences in medication use.
In Aim 2, we will develop a dynamic risk assessment algorithm for AD using longitudinal vascular markers and compare the performance of this new algorithm with existing AD assessment risk scores.
In Aim 3 we will identify longitudinal vascular characteristics associated with conversion to dementia in participants with mild cognitive impairment (MCI).
In Aim 4, we will examine the association between longitudinal vascular marker trajectories and longitudinal cognitive function using functional regression models to determine how changes in the vascular markers are related to changes in cognitive function.
The proposed project will be the first study to explore a dynamic relationship between multiple longitudinal vascular measures and AD in an African American cohort. Results from this study can provide a more accurate AD risk assessment method based on data already routinely collected in clinical practices. Our results may also lead to better strategies for potential interventions in elderly individuals.
|Hendrie, Hugh C; Zheng, Mengjie; Li, Wei et al. (2017) Glucose level decline precedes dementia in elderly African Americans with diabetes. Alzheimers Dement 13:111-118|
|Yang, Lili; Yu, Menggang; Gao, Sujuan (2016) Joint Models for Multiple Longitudinal Processes and Time-to-event Outcome. J Stat Comput Simul 86:3682-3700|
|Yang, Lili; Yu, Menggang; Gao, Sujuan (2016) Prediction of coronary artery disease risk based on multiple longitudinal biomarkers. Stat Med 35:1299-314|
|Lai, Dongbing; Xu, Huiping; Koller, Daniel et al. (2016) A MULTIVARIATE FINITE MIXTURE LATENT TRAJECTORY MODEL WITH APPLICATION TO DEMENTIA STUDIES. J Appl Stat 43:2503-2523|